Y+ problems for a wind turbine blades

Elio · September 11, 2013, 19:16

I am modeling a 3D case of a vertical axis wind turbine. I calculated the first layer (which was 3e-5 m)however i got a y+ value of 708 after 1 complete rotation of the turbine!!! It's a complete mess. I need a y+ value of 1. I already have 11 million elements. Do you think the solution has not converged completely yet? Or is there any other way? Please advise
In addition one rotation is taking around 2 days to get a solution. So I need a solution that could get me going faster.
Many thanks

ghorrocks · September 12, 2013, 18:16

Why do you need y+ of 1? If your flow turbulent then a y+>>1 is OK and you can use wall functions. Wall functions are accurate when correctly applied.

The best way to determine what mesh you need is with a mesh sensitivity study. You need to do this to see what mesh size you need. At the moment you are guessing.

To make a simulation go faster get the thing running in parallel. That is why supercomputers are built to run CFD simulations - they are big and require serious computing power.

Elio · September 12, 2013, 18:19

Sir I am using 7 parallel CPUS with 98 GB rams. However, I need to catch the laminar layer on the blades walls. It's a transient model so things are changing and it's getting all mixed up. If you have another advice please do help

ghorrocks · September 12, 2013, 18:29

Why do you need to catch the laminar BL on the blades? What Re are the blade running at?

You are doing well to get a simulation like this to run in 2 days. I think by the time you refine the mesh it will be much slower. So you will need faster and more computers. But before you do any of that, have a careful think:
* Do I need to model the whole thing in transient? Can a simplification give me enough information?
* Do I know what mesh size I need? Do a mesh sensitivity study on a small simple section before doing the whole thing.
* Likewise for time step size and convergence criterion.
* Why do you need to model the laminar BL? Be sure you actually need this before committing to it - it is going to make your simulation many times bigger.
* If you are talking about laminar BL then do you need a turbulence transition model?

September 11, 2013, 19:16	Y+ problems for a wind turbine blades	#1
Elio New Member Join Date: Dec 2012 Posts: 12 Rep Power: 13	I am modeling a 3D case of a vertical axis wind turbine. I calculated the first layer (which was 3e-5 m)however i got a y+ value of 708 after 1 complete rotation of the turbine!!! It's a complete mess. I need a y+ value of 1. I already have 11 million elements. Do you think the solution has not converged completely yet? Or is there any other way? Please advise In addition one rotation is taking around 2 days to get a solution. So I need a solution that could get me going faster. Many thanks

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[ICEM] Problems with coedge curves and surfaces	tommymoose	ANSYS Meshing & Geometry	6	December 1, 2020 11:12
analysis of wind turbine blades	sharuk	Main CFD Forum	7	December 9, 2014 18:59
Moving reference frame in wind turbine	kongl1986	FLUENT	0	March 30, 2013 10:50
Vertical Axis Wind Turbine Rotating Domain Problems	TWaung	CFX	4	May 1, 2012 03:14
I want to simulate a wind turbine in a yaw position ( lateral wind)	mohammad	FLOW-3D	1	January 17, 2011 01:41

September 12, 2013, 18:16		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Why do you need y+ of 1? If your flow turbulent then a y+>>1 is OK and you can use wall functions. Wall functions are accurate when correctly applied. The best way to determine what mesh you need is with a mesh sensitivity study. You need to do this to see what mesh size you need. At the moment you are guessing. To make a simulation go faster get the thing running in parallel. That is why supercomputers are built to run CFD simulations - they are big and require serious computing power.

September 12, 2013, 18:19		#3
Elio New Member Join Date: Dec 2012 Posts: 12 Rep Power: 13	Sir I am using 7 parallel CPUS with 98 GB rams. However, I need to catch the laminar layer on the blades walls. It's a transient model so things are changing and it's getting all mixed up. If you have another advice please do help

September 12, 2013, 18:29		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Why do you need to catch the laminar BL on the blades? What Re are the blade running at? You are doing well to get a simulation like this to run in 2 days. I think by the time you refine the mesh it will be much slower. So you will need faster and more computers. But before you do any of that, have a careful think: * Do I need to model the whole thing in transient? Can a simplification give me enough information? * Do I know what mesh size I need? Do a mesh sensitivity study on a small simple section before doing the whole thing. * Likewise for time step size and convergence criterion. * Why do you need to model the laminar BL? Be sure you actually need this before committing to it - it is going to make your simulation many times bigger. * If you are talking about laminar BL then do you need a turbulence transition model?